Install
Works with Claude Code, Claude apps, Codex, Cursor, Gemini CLI, GitHub Copilot, VS Code, OpenCode, Goose, Windsurf. Read the repository before you give any tool access to your data.
npx agentmarket install skill/sushilti80-spec-recipes
Asks which of your tools to add it to (Claude Code, Codex, Cursor, VS Code, Windsurf, Gemini CLI, Claude Desktop) and shows the change first. See https://agentmart.dev/cli.
Get the weekly diff
The weekly diff: one email a week with new agent tools, notable updates and grade changes, plus security alerts for tools you follow. Double opt-in, one-click unsubscribe.
About
Use cases: data, productivity
Reviews
Write a reviewNo reviews yet. Installed it? Reviews come from verified installers only. How reviews work
Permissions
What this skill can do and what it asks for, from its SKILL.md. Checked Oct 25, 2026.
- Tools
- No tools found
- Secrets
- None asked for
- Install scripts
- None found
Network hosts it talks to 1
codeload.github.com· found in install command
Files it touches
~/.claude/skills/spec-recipes
What's inside
- Skill name
spec-recipes- Description
- Need-based workflow recipes for eng-orchestrator. Identify need first, build minimal plan from recipe×tier matrix, add agents only when checklist flags risk, adapt mid-flight. Alias capability = greenfield-feature. /spec-pipeline defers here.
- License
- MIT
- Instructions
- 75 lines in SKILL.md · view on GitHub
Files (1)
- SKILL.md
SKILL.md (first lines)
# Spec recipes Full definitions: `SPECFORGE_HOME/ENGINEERING-RECIPES.md` (§0 + matrix). Orchestrator: `agents/eng-orchestrator.md`. Playbook: `SPECFORGE_HOME/ENGINEERING-PLAYBOOK.md`. **No default production recipe.** Classify need → smallest recipe × tier → matrix R/O/—. ## Rules 1. Build from **minimal + matrix R/O/—**, not maximal ASCII pipelines. 2. **Reclassify** when evidence changes; announce before continuing. 3. HANDOFF must include: `Recipe`, `Tier`, `Need summary`, `Plan (R/O/skipped)`, `Adapt watchers`, `parent_REQ` when relevant. 4. Conflict with other docs: **recipes omit wins** unless checklist flags risk. 5. Human owns `APPROVED` / BUG `Human ACK` / override / Critical & Blocking-drift waivers. Loops ≤2 rounds. 6. **Two-pass:** planning agents only → Phase `awaiting-human` (user edits disk) → implementation agents. Chat LGTM ≠ APPROVED. `READY_FOR_APPROVAL` is HANDOFF-only. ## Quick picker (after need checklist) | Need | Recipe | |------|--------| | Spike / should we | Tier 0 note or `advisory-only` | | Docs only | `docs-touch` | | Vendor sync | `vendor-sync` | | Specs only | `spec-only` | | New product | `new-application` | | New user-facing capability | `greenfield-feature` (`capability`) | | Defect, normal | `bug-fix` | | Prod-urgent | `hotfix` | | Deps/refactor | `maintenance` | | IaC/CI/env | `infra-change` | | CVE / security | `security-patch` | ## Gates (only those implied by **R** agents) | Gate | When required | |------|----------------| | G-plan | Planning agents finished; files may still be DRAFT/OPEN | | G-human | User set disk `Status: APPROVED` or BUG `Human ACK: APPROVED` before coding | | G1 REQ user-APPROVED | Before architect / implementers per matrix | | G2 ARCH user-APPROVED | When ARCH is **R** for this tier | | G3 tests + Critical-clear | Before verifier when test/review are **R** | | G4 verify + Blocking drift clear | Before DONE when verifier/guardian are **R** | ## Minimal examples (not ceilings) - **Capability Tier 1:** REQ → **user APPROVED** → implementer → test-runner → verifier - **Bug-fix Tier 1:** debugger (BUG) → implementer → test-runner → code-reviewer → verifier - **Hotfix:** debugger → implementer → test-runner → verifier (+ security if adjacent) Full ASCII of every agent is a **Tier 3 ceiling**, not the default. See recipes matrix. ## Principle 8 After each gate: update specs + memory (include **agents_planned**) → optional GATE file → **fresh subagent** with paths only. Never attach chat summaries. ## Invoke ``` /eng-orchestrator — diagnose need, then minimal recipe × tier plan ``` Alias command `/spec-pipeline` opens this same skill (entry cheat sheet). ## Related skills | Skill | Role | |-------|------| | `spec-req-author` / `spec-arch-author` | Write DRAFT specs | | `spec-challenger` | Adversarial review (capped) | | `spec-handoff` | End-of-phase HANDOFF | | `spec-verifier` / `spec-guardian-drift` | Verify / drift | | `spec-token-budget` / `spec-advisory` / `spec-vendor-sync` | Token / meta | | `ponytail` / `ponytail-review` | Minimal code / Gate 3 bloat |
From sushilti80/specforge-engineering-team, MIT.
More from sushilti80/specforge-engineering-team
- 0.524score
ui-a11y
sushilti80
Accessibility checklist for interactive UI: labels, keyboard, focus, contrast, and ARIA for overlays.
SkillBFreshness: BClaude Code +9 - 0.524score
spec-advisory
sushilti80
Readonly advisory mode — architecture review, compare options, feasibility, "should we", "critical review", "worth it", "review only".
SkillBFreshness: BClaude Code +9 - 0.524score
ui-forms
sushilti80
Build working forms: schema validation, field errors, submit/pending/disabled states, and accessible labels.
SkillBFreshness: BClaude Code +9 - 0.521score
spec-agent-audit
sushilti80
Readonly audit of agent definition files in agents/ for memory & self-learning patterns, debugging patterns, description alignment, and contradicting…
SkillBFreshness: BClaude Code +9 - 0.521score
spec-arch-author
sushilti80
Write ARCH specs and ADRs from user-APPROVED REQ files.
SkillBFreshness: BClaude Code +9 - 0.521score
spec-challenger
sushilti80
Adversarial review of REQ or ARCH (or consequential ADR) with capped rounds.
SkillBFreshness: BClaude Code +9 - 0.521score
spec-guardian-drift
sushilti80
Readonly compare .specs/ to codebase and flag drift.
SkillBFreshness: BClaude Code +9 - 0.521score
spec-handoff
sushilti80
Standard HANDOFF block and checkpoint files for spec-driven pipeline.
SkillBFreshness: BClaude Code +9 - 0.521score
spec-agent-memory
sushilti80
Read and update project agent memory in .agents/memory/.
SkillBFreshness: BClaude Code +9 - 0.521score
spec-release-metrics
sushilti80
Record release efficiency against need-based plans: planned vs run agents, estimated tokens (minimal mode), human gates, quality KPIs.
SkillBFreshness: BClaude Code +9 - 0.521score
spec-req-author
sushilti80
Write or update REQ specs in .specs/requirements/.
SkillBFreshness: BClaude Code +9 - 0.521score
spec-token-budget
sushilti80
Token and output budgets by task type.
SkillBFreshness: BClaude Code +9
Signals
- Stars
- 0
- Forks
- 0
- Open issues
- 0
- Last commit (this folder)
- Aug 6, 2026 (2 months ago)
- Last release
- Aug 6, 2026 (2 months ago)
From GitHub (fetched Oct 24, 2026).
Details
- License
- MIT
- First seen
- Oct 24, 2026
- Last checked
- Oct 24, 2026
- Source
- GitHubgithub-graphql
More from sushilti80
- 0.524score
ui-a11y
sushilti80
Accessibility checklist for interactive UI: labels, keyboard, focus, contrast, and ARIA for overlays.
SkillBFreshness: BClaude Code +9 - 0.524score
ui-forms
sushilti80
Build working forms: schema validation, field errors, submit/pending/disabled states, and accessible labels.
SkillBFreshness: BClaude Code +9 - 0.524score
spec-advisory
sushilti80
Readonly advisory mode — architecture review, compare options, feasibility, "should we", "critical review", "worth it", "review only".
SkillBFreshness: BClaude Code +9 - 0.521score
spec-challenger
sushilti80
Adversarial review of REQ or ARCH (or consequential ADR) with capped rounds.
SkillBFreshness: BClaude Code +9 - 0.521score
spec-arch-author
sushilti80
Write ARCH specs and ADRs from user-APPROVED REQ files.
SkillBFreshness: BClaude Code +9 - 0.521score
spec-agent-memory
sushilti80
Read and update project agent memory in .agents/memory/.
SkillBFreshness: BClaude Code +9
Cite this page
spec-recipes skill. Agent Mart. https://agentmart.dev/skills/sushilti80-spec-recipes (last updated 2026-10-24)